actions-debugging
Use when a GitHub Actions workflow fails — diagnose the run log, identify the root cause, and apply a targeted fix
Install
npx skills add https://github.com/drvoss/everything-copilot-cli/tree/main/skills/copilot-exclusive/actions-debugging
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install drvoss-everything-copilot-cli@llmmart
git clone https://github.com/drvoss/everything-copilot-cli.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole drvoss/everything-copilot-cli collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
GitHub Actions Debugging
Why This is Copilot-Exclusive
GitHub access is built into Copilot CLI with no extra setup: the built-in GitHub MCP server (full
tool set when gh is absent, or when restored with the flags in the note below) or gh, which
the model is expected to use when it is installed. Either way you can list workflows, inspect run details, download job logs,
and analyze failures in one conversation.
What is exclusive is the built-in GitHub MCP server, which needs no token or extension setup; when gh is installed, the model mostly uses gh, which any agent with a shell can also run.
Note: When
ghis on PATH, Copilot CLI omits the Actions MCP tools (actions_list,actions_get,get_job_logs) and the model is expected to useghinstead, e.g.gh run list --workflow <file>,gh run view <run-id> --log-failed,gh run view --job <job-id> --log. To call the MCP tools, start the session with--add-github-mcp-tool=<name>for every GitHub MCP tool you need (it replaces the default set, so include default ones you still use), or--enable-all-github-mcp-tools, which also enables write tools.
When to Use
- A CI pipeline fails and you need to understand why — fast
- Investigating flaky tests across multiple workflow runs
- Auditing workflow run times and usage patterns
- Debugging deployment pipeline failures in staging/production
- Comparing successful vs failed runs to find regressions
Workflow
1. List Workflows in a Repository
With gh on PATH (default), e.g. gh workflow list --repo my-org/my-app. When the MCP tools are exposed:
Tool: github-mcp-server-actions_list
method: "list_workflows"
owner: "my-org"
repo: "my-app"
2. List Recent Runs for a Workflow
With gh on PATH (default), e.g. gh run list --repo my-org/my-app --workflow ci.yml --branch main --status completed. When the MCP tools are exposed:
Tool: github-mcp-server-actions_list
method: "list_workflow_runs"
owner: "my-org"
repo: "my-app"
resource_id: "ci.yml"
workflow_runs_filter: { "status": "completed", "branch": "main" }
3. Get Failed Job Logs
This is the killer feature — pull logs directly for analysis.
With gh on PATH (default), e.g. gh run view 12345678 --repo my-org/my-app --log-failed. When the MCP tools are exposed:
Tool: github-mcp-server-get_job_logs
owner: "my-org"
repo: "my-app"
run_id: 12345678
failed_only: true
return_content: true
tail_lines: 200
4. Inspect a Specific Job
With gh on PATH (default), e.g. gh run view --job 98765432 --repo my-org/my-app. When the MCP tools are exposed:
Tool: github-mcp-server-actions_get
method: "get_workflow_job"
owner: "my-org"
repo: "my-app"
resource_id: "98765432"
5. Check Run Usage and Timing
With gh on PATH (default), e.g. gh api repos/my-org/my-app/actions/runs/12345678/timing. When the MCP tools are exposed:
Tool: github-mcp-server-actions_get
method: "get_workflow_run_usage"
owner: "my-org"
repo: "my-app"
resource_id: "12345678"
Examples
"Why Did My CI Fail?"
"My latest push to feature/auth-refactor failed CI. Show me what went wrong."
Copilot workflow:
gh run list --branch feature/auth-refactor(when exposed:list_workflow_runsfiltered by that branch)- Takes the most recent failed run
gh run view <run-id>to find which job(s) failed (when exposed:list_workflow_jobs)gh run view <run-id> --log-failedto pull the error output (when exposed:get_job_logswithfailed_only: true)- Analyzes the logs, identifies the root cause, suggests a fix
All in one conversational turn — no browser tabs needed.
Flaky Test Investigation
"The 'integration-tests' workflow has been flaky on main this week. Show me the failure pattern."
With gh on PATH (default), e.g. gh run list --repo my-org/my-app --workflow integration-tests.yml --branch main --limit 20. When the MCP tools are exposed:
Tool: github-mcp-server-actions_list
method: "list_workflow_runs"
owner: "my-org"
repo: "my-app"
resource_id: "integration-tests.yml"
workflow_runs_filter: { "branch": "main" }
per_page: 20
Copilot lists the last 20 runs, identifies which failed, pulls logs for each failure, and looks for common error patterns (e.g., timeout, race condition, external service unavailability).
Deployment Pipeline Audit
"Show me all production deployment runs this month and flag any that took longer than 15 minutes."
Copilot lists runs, checks usage/timing for each, and highlights anomalies.
Fix and Verify Loop
- Copilot reads the failure logs
- Identifies the broken test or build step
- Makes the fix in your local codebase
- You push the fix
- Copilot monitors the new run via
gh run list/gh run watch <run-id>(when exposed:list_workflow_runs) - Confirms the fix worked or continues debugging
Tips
- Use
failed_only: true(withgh:--log-failed): Skip successful job logs and go straight to the failures. Saves context window space. - Tail the logs (MCP only): Use
tail_lines: 100-300to focus on the end of the log where errors typically appear. Withgh,gh run view <run-id> --log-failedalready limits output to the failed steps. - Parallel job analysis: When a run has multiple failed jobs, call
get_job_logs(withgh:gh run view --job <job-id> --log) on each in parallel. - Compare runs: Pull logs from both a passing and failing run to diff them and isolate what changed.
- Combine with PR tools: After fixing a CI failure, use
pull_request_readwithget_check_runs(withgh:gh pr checks <n>) to verify the fix passes on the PR. - Track flaky tests in SQL: Store failure patterns in the session database to build a flakiness report over time.
Files (everything-copilot-cli)
-
SKILL.md 6.1 KB
--- name: actions-debugging license: MIT description: Use when a GitHub Actions workflow fails — diagnose the run log, identify the root cause, and apply a targeted fix metadata: category: copilot-exclusive copilot_feature: "Built-in GitHub MCP server (actions_list, actions_get, get_job_logs)" --- # GitHub Actions Debugging ## Why This is Copilot-Exclusive GitHub access is built into Copilot CLI with no extra setup: the built-in GitHub MCP server (full tool set when `gh` is absent, or when restored with the flags in the note below) or `gh`, which the model is expected to use when it is installed. Either way you can list workflows, inspect run details, download job logs, and analyze failures in one conversation. What is exclusive is the built-in GitHub MCP server, which needs no token or extension setup; when `gh` is installed, the model mostly uses `gh`, which any agent with a shell can also run. > **Note:** When `gh` is on PATH, Copilot CLI omits the Actions MCP tools (`actions_list`, `actions_get`, `get_job_logs`) and the model is expected to use `gh` instead, > e.g. `gh run list --workflow <file>`, `gh run view <run-id> --log-failed`, `gh run view --job <job-id> --log`. To call the MCP tools, > start the session with `--add-github-mcp-tool=<name>` for every GitHub MCP tool you need (it replaces the default set, so include default ones you still use), or `--enable-all-github-mcp-tools`, which also enables write tools. ## When to Use - A CI pipeline fails and you need to understand why — fast - Investigating flaky tests across multiple workflow runs - Auditing workflow run times and usage patterns - Debugging deployment pipeline failures in staging/production - Comparing successful vs failed runs to find regressions ## Workflow ### 1. List Workflows in a Repository With `gh` on PATH (default), e.g. `gh workflow list --repo my-org/my-app`. When the MCP tools are exposed: ```text Tool: github-mcp-server-actions_list method: "list_workflows" owner: "my-org" repo: "my-app" ``` ### 2. List Recent Runs for a Workflow With `gh` on PATH (default), e.g. `gh run list --repo my-org/my-app --workflow ci.yml --branch main --status completed`. When the MCP tools are exposed: ```text Tool: github-mcp-server-actions_list method: "list_workflow_runs" owner: "my-org" repo: "my-app" resource_id: "ci.yml" workflow_runs_filter: { "status": "completed", "branch": "main" } ``` ### 3. Get Failed Job Logs This is the killer feature — pull logs directly for analysis. With `gh` on PATH (default), e.g. `gh run view 12345678 --repo my-org/my-app --log-failed`. When the MCP tools are exposed: ```text Tool: github-mcp-server-get_job_logs owner: "my-org" repo: "my-app" run_id: 12345678 failed_only: true return_content: true tail_lines: 200 ``` ### 4. Inspect a Specific Job With `gh` on PATH (default), e.g. `gh run view --job 98765432 --repo my-org/my-app`. When the MCP tools are exposed: ```text Tool: github-mcp-server-actions_get method: "get_workflow_job" owner: "my-org" repo: "my-app" resource_id: "98765432" ``` ### 5. Check Run Usage and Timing With `gh` on PATH (default), e.g. `gh api repos/my-org/my-app/actions/runs/12345678/timing`. When the MCP tools are exposed: ```text Tool: github-mcp-server-actions_get method: "get_workflow_run_usage" owner: "my-org" repo: "my-app" resource_id: "12345678" ``` ## Examples ### "Why Did My CI Fail?" > "My latest push to feature/auth-refactor failed CI. Show me what went wrong." Copilot workflow: 1. `gh run list --branch feature/auth-refactor` (when exposed: `list_workflow_runs` filtered by that branch) 2. Takes the most recent failed run 3. `gh run view <run-id>` to find which job(s) failed (when exposed: `list_workflow_jobs`) 4. `gh run view <run-id> --log-failed` to pull the error output (when exposed: `get_job_logs` with `failed_only: true`) 5. Analyzes the logs, identifies the root cause, suggests a fix All in one conversational turn — no browser tabs needed. ### Flaky Test Investigation > "The 'integration-tests' workflow has been flaky on main this week. > Show me the failure pattern." With `gh` on PATH (default), e.g. `gh run list --repo my-org/my-app --workflow integration-tests.yml --branch main --limit 20`. When the MCP tools are exposed: ```text Tool: github-mcp-server-actions_list method: "list_workflow_runs" owner: "my-org" repo: "my-app" resource_id: "integration-tests.yml" workflow_runs_filter: { "branch": "main" } per_page: 20 ``` Copilot lists the last 20 runs, identifies which failed, pulls logs for each failure, and looks for common error patterns (e.g., timeout, race condition, external service unavailability). ### Deployment Pipeline Audit > "Show me all production deployment runs this month and flag any that took > longer than 15 minutes." Copilot lists runs, checks usage/timing for each, and highlights anomalies. ### Fix and Verify Loop 1. Copilot reads the failure logs 2. Identifies the broken test or build step 3. Makes the fix in your local codebase 4. You push the fix 5. Copilot monitors the new run via `gh run list` / `gh run watch <run-id>` (when exposed: `list_workflow_runs`) 6. Confirms the fix worked or continues debugging ## Tips - **Use `failed_only: true`** (with `gh`: `--log-failed`): Skip successful job logs and go straight to the failures. Saves context window space. - **Tail the logs** (MCP only): Use `tail_lines: 100-300` to focus on the end of the log where errors typically appear. With `gh`, `gh run view <run-id> --log-failed` already limits output to the failed steps. - **Parallel job analysis**: When a run has multiple failed jobs, call `get_job_logs` (with `gh`: `gh run view --job <job-id> --log`) on each in parallel. - **Compare runs**: Pull logs from both a passing and failing run to diff them and isolate what changed. - **Combine with PR tools**: After fixing a CI failure, use `pull_request_read` with `get_check_runs` (with `gh`: `gh pr checks <n>`) to verify the fix passes on the PR. - **Track flaky tests in SQL**: Store failure patterns in the session database to build a flakiness report over time.
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.